Mutation Profiling in Uterine Carcinosarcoma / Malignant Mixed Mullerian Tumors

An ESUN Study Report
By Melissa McConechy; David Huntsman, MD;
Cheng-Han Lee, MD, PhD and Blaise Clarke, MD

Introduction

Carcinosarcoma of the uterus (also known as malignant mixed Mullerian tumor, MMMT) is a highly aggressive form of uterine cancer.^1-3 Even though it constitutes about 3-4% of uterine malignancy overall, it accounts for a disproportionate percentage of mortality associated with uterine malignancy. As its name implies, this is a biphasic (two-component) tumor which contains an admixture of carcinoma (cancer showing epithelial differentiation) and sarcoma (cancer showing mesenchymal differentiation) components. In the great majority of uterine carcinosarcoma, both the carcinoma and the sarcoma components are histologically high-grade. While a number of groups have previously demonstrated that the carcinoma and the sarcoma components represent the same disease based on X-chromosome inactivation pattern and TP53 hotspot mutation sequencing studies,^4,5 it is not certain whether uterine carcinosarcoma starts out as carcinoma that undergoes sarcomatous transformation or as sarcoma that undergoes epithelial differentiation (similar to other biphasic sarcomas like synovial sarcoma).

We have previously demonstrated the presence of at least two major genetic types of uterine carcinosarcoma, each with its distinct gene mutation profiles (endometrioid carcinoma-like profile and serous carcinoma-like profile; see Figure 1),⁶ and we speculated as others have that uterine carcinosarcoma arises through sarcomatous transformation/transdifferentiation of a pre-existing uterine carcinoma, hence maintaining the mutation profiles of the pre-existing carcinoma.^3,7

Carcinosarcomas or malignant mullerian mixed tumors (MMMT) contain malignant epithelial (carcinoma) and mesenchymal (sarcoma) components. These tumors can arise in any organ of the female genital tract such as vulva, vagina, cervix, endometrium, ovary or fallopian tube. Endometrium is the most common site. These tumors tend to occur in post-menopausal women although rarely are reported in younger women. Clinically, patients with uterine MMMT present with abnormal vaginal bleeding or pelvic or abdominal pain. Uterine carcinosarcomas generally have a worse prognosis compared to uterine carcinomas with a five year survival rate ranging from 5% to 40%. The most important prognostic factor is tumor stage.

The sarcomatous component may be either homologous or heterologous. The homologous component is usually non-specific high grade sarcoma in keeping with undifferentiated endometrial sarcoma. Heterologous elements include rhabdomyosarcoma, chondrosarcoma or osteosarcoma. Uterine carcinosarcomas follow the same staging system as uterine carcinomas.

Study Aims

1. To more comprehensively characterize the mutation profiles of uterine carcinosarcomas through targeted sequencing.
2. To compare the mutation profiles of the carcinoma and the sarcoma components in uterine carcinosarcoma to determine the relationship between the two components and the directionality of tumor progress.
3. To compare the mutation profiles of primary uterine tumor (both carcinoma and sarcoma components) and metastatic tumor (peritoneal metastasis) to determine whether potentially targetable mutation(s) is present in all tumor sites.

Results

We performed a comprehensive mutation analysis of 25 cancer genes that have been previously implicated in the oncogenesis of different uterine carcinomas and uterine sarcomas.^6,8-13 These included ABCC9, AKT1, AKT2, AKT3, ARID1A, CCND1, CHD4, CSMD3, CTCF, CTNNB1, EP300, FBXW7, FGFR2, MAP3K4, MED12, PIK3CA, PIK3R1, PIK3R2, POLE, PPP2R1A, PTEN, SPOP, TP53, TSPYL2 and ZFHX3. The Illumina custom TruSeq amplicon panel was designed using Illumina’s DesignStudio and includes 1519 amplicons that cover nearly all exons of these 25 genes. We applied these analyses on a series of uterine carcinosarcomas with available frozen tumor tissue, formalin-fixed paraffin-embedded (FFPE) tumor tissue and normal tissue (Buffy coat). Libraries were constructed with genomic DNA extracted from matched tumor and normal tissue, pooled and run on the MiSeq (Illumina). In tumors with spatially distinct carcinoma and sarcoma components, we extracted DNA from the distinct carcinoma-only and sarcoma-only components, and analyzed these components separately. We also extracted DNA from metastatic tumor foci for comparison. All somatic non-synonymous mutations were manually checked in bam files using Integrated Genome Viewer.

Uterine carcinosarcomas exhibit mutation profiles similar to either uterine endometrioid-type or serous-type carcinomas.

With this set of 25 cancer genes, we identified the presence of somatic mutations (missense and non-sense point mutations, small insertions and deletions) in all primary uterine carcinosarcomas examined (n=30). The number of genes mutated ranged from 1 to 14 genes per tumor sample, with a median of 4 genes mutated per tumor (with a mutation variant frequency of at least 10% in the tumor sample). Based on the mutation profiles, we confirmed our prior observation that most uterine carcinosarcomas exhibit a mutation pattern that resembles that of either endometrioid-type carcinoma or serous-type carcinoma of the uterus. The endometrioid carcinoma-like mutation profile consisted of a clustering of PTEN, ARID1A, PIK3R1 and POLE mutations, while the serous carcinoma-like mutation profile consisted of a clustering of TP53, PPP2R1A, EP300 and FBXW7 mutations.

Endometrial cancer is not a single disease but is composed of multiple different subtypes which have differing risk factors, precursor lesions, genetic changes, treatment options and clinical outcomes. Typically the subtypes include endometrioid, serous, clear cell, undifferentiated and mixed carcinomas (those composed of more than one subtype)

Previously, targeted sequencing of exons from 9 genes: ARID1A, PPP2R1A, PTEN, PIK3CA, KRAS, CTNNB1, TP53, BRAF, and PPP2R5C in a large cohort of endometrial carcinomas highlighted distinct mutation profiles between subtypes. Grade 3 endometrioid carcinomas have significantly different frequencies of PTEN and TP53 mutations compared to low-grade counterparts. Uterine serous carcinoma and grade 3 endometrioid carcinomas have significantly different frequencies of mutations in PTEN, ARID1A, PPP2R1A, TP53, and CTNNB1. The genomic profile of carcinosarcomas suggests two distinct mutation profiles: endometrioid-type (PTEN, PIK3CA, ARID1A, KRAS mutations) and serous-type (TP53 and PPP2R1A mutations).⁶

Uterine carcinosarcoma occurs through sarcomatous transformation/transdifferentiation from uterine carcinomas

In nearly all the cases where the carcinoma and the sarcoma components were separately analyzed, we found that these components shared identical mutation(s) in at least one gene (1 to 11 genes, median of 2 genes), indicating a common ancestral origin (i.e. both components represent parts of the same tumor). There were a significantly greater number of genes mutated in the sarcoma component compared to the carcinoma component, indicating a carcinoma to sarcoma directionality of tumor progression. These findings suggest that the sarcoma component arose from pre-existing endometrial carcinoma of either endometrioid-genotype or serous–genotype (Figure 1). In a subset of uterine carcinomas that exhibits a mutation profile in keeping with an origin from endometrioid-type carcinoma, we observed high-frequency TP53 mutation in only the sarcoma component but not the carcinoma component, thus implicating TP53 mutation as being a candidate genomic event underlying the sarcomatous transformation phenomenon in endometrioid-genotype tumors.

There have been several proposed mechanisms by which carcinosarcomas may develop:

1. The collision theory posits that the two elements, the carcinoma and the sarcoma, arise independently and collide to appear as a single tumor
2. The combination theory claims that both tumor types arise from a single stem cell/progenitor cell that then differentiates into both sarcoma and carcinoma
3. The conversion theory maintains that the sarcomatous elements evolves from the carcinoma

Genomic evidence of PI3K pathway dysregulation is common in uterine carcinosarcoma, and is present in both primary and metastatic tumor foci, hence representing a compelling therapeutic target

We observed frequent genomic evidence of PI3K pathway dysregulation in uterine carcinosarcomas. Mutations involving PI3K pathway regulatory genes were identified in 60% of all primary tumors, most commonly involving PIK3CA (50% of the tumors), followed by PTEN (30% of the tumors) and PIK3R1 (30% of the tumors). More importantly, these PI3K pathway gene mutations were shared by not only the carcinoma and the sarcoma components of the primary tumors, but also by the corresponding metastatic tumors (Figure 2). These findings indicate that PI3K pathway mutations occur relatively early on in the development of uterine carcinosarcoma (i.e. prior to sarcomatous transformation/transdifferentiation, and prior to tumor metastasis), and represent a common therapeutic target in the majority of uterine carcinosarcomas.

In a recent study, predictive biomarkers for single agent mTOR inhibitors were sought by profiling archival tumor samples from patients with recurrent or metastatic endometrial cancer who participated in phase 2 trials investigating single-agent mTOR inhibitors. Analyses included mutational profiling using the OncoCarta Panel version 1.0 and immunohistochemical expression of the tumor suppressor gene PTEN and stathmin, a marker of PI3K activation. No correlation was observed between biomarkers and tumor response or progression.¹⁴

Summary

Our findings provide further support that uterine carcinosarcomas arise through transdifferentiation of uterine carcinoma into sarcoma, and maintain the mutation profiles of the pre-existing uterine carcinoma (endometrioid carcinoma-like or serous carcinoma-like). We observed frequent PI3K pathway mutations in uterine carcinosarcoma which are acquired prior to tumor transdifferentiation and metastasis. These findings provide biologically compelling support for the use of therapy that target PI3K signaling and/or downstream pathway in the majority of uterine carcinosarcoma. Further functional studies that evaluate the effects of therapies targeting PI3K and/or downstream pathway in biologically representative experimental models of uterine carcinosarcoma is needed to determine the efficacy of this therapeutic strategy, both as a single therapy and when combined with conventional therapies.

Introduction

Uterine carcinosarcoma (formerly called malignant mixed mullerian tumor) is a rare tumor of the gynecologic tract. It is classified as a mixed epithelial and mesenchymal tumor of the uterus in the 2003 World Health Organization classification. It accounts for <5% of uterine malignancies and typically arises in the uterine corpus or in the cervix.¹ Rare cases arising in the ovary and fallopian tube have also been described.^2,3 It is an aggressive neoplasm that has an extremely poor prognosis: 5-year survival is about 35%. Therapeutic options are limited and no specific targeted therapy is available.

Who gets uterine carcinosarcoma?

Uterine carcinosarcoma affect postmenopausal women, with a higher incidence among African American women. Although it is a rare tumor with an incidence of 2 per 100,000 women per year, it accounts for a disproportionate percentage of uterine cancer-associated deaths. Risk factors associated with this disease are similar to those encountered in endometrial carcinoma, and include obesity and prolonged estrogen exposure. Some cases have been associated with prior radiotherapy.⁴

How do patient's present with this disease?

The clinical symptoms are essentially non-specific. The most common symptom is abnormal vaginal bleeding, or a bloody or watery discharge. Some patients may present with a mass or a bulky tumor which protrudes through the uterine cervix into the vagina. In addition to this, patients may have non-specific systemic symptoms such as fatigue. These are aggressive neoplasms and in 60% of patients the disease will have extended beyond the confines of the uterus at the time of presentation.

How is uterine carcinosarcoma diagnosed?

Once patients present with such symptoms they will undergo a uterine biopsy, and a pathologist will review the biopsy. Occasionally both components may not be sampled on biopsy and the final diagnosis will be rendered on the hysterectomy specimen. These are biphasic tumors showing an admixture of high grade carcinoma and sarcoma, although the exact volume and subtype of each component will vary (see Figures 1 & 2). The epithelial component may be endometrioid, serous and rarely clear cell or undifferentiated carcinoma. The sarcomatous component may be either homologous or heterologous. Homologous refers to the fact that the sarcomatous component is one which is native to the gynecologic tract and these typically include undifferentiated sarcoma, fibrosarcoma or leiomyosarcoma. Heterologous mesenchymal elements include rhabdomyosarcoma and chondrosarcoma with rare cases of liposarcoma, osteosarcoma and neurectodermal differentiation having been reported.⁴

How do we treat patients with uterine carcinosarcoma?

These patients will be managed with complete surgical staging undergoing a total abdominal hysterectomy, bilateral salpingo-oophrectomy, pelvic and para-aortic lymphadenectomy, omentectomy biopsies of peritoneal surfaces and tumor debulking.¹ If the tumor has spread widely at the time of diagnosis, surgical excision may not be possible. In addition to surgery, adjuvant radiation therapy and chemotherapy may be used. Radiation therapy reduces local pelvic recurrence but it is unclear whether there is improvement in overall survival.^{6,7,8 9} Since many of these patients will have disseminated disease and recur in a metastatic site, chemotherapy is used but there is uncertainty as to the best treatment.^10,11,12,13 Multimodality therapy is often used with patients receiving both chemotherapy and radiation therapy.

The Clinical Problem

Despite extensive surgery and multimodality adjuvant therapy, these tumors have an extremely poor prognosis: 5-year survival is about 35%. There is no standard second line therapy and no specific targeted therapy is available. Traditional chemotherapies also have multiple side effects since they are targeting dividing cells, including normal cells. It is hoped that targeted therapeutics would be more effective in treating the tumor and cause less side effects. Given the limitations of surgery and radiotherapy, the only realistic hope for improving outcomes is improving chemotherapy.

Difficulties in Conducting Uterine Carcinosarcoma Research

Given the relatively low frequency of this disease within the population, intelligent trial design and drug selection is crucial when seeking to improve the outcome for women with this disease. Obtaining additional biological information that can guide selection of agents for investigation in this underserviced patient population is essential.

The Origin of Uterine Carcinosarcoma

A number of theories as to the cell of origin of such tumors has been proposed.¹⁴ The collision theory posits that these are biclonal synchronous tumors arising from separate tumour-initiating cells (TICs) that subsequently merge. The conversion theory posits that these are monoclonal tumors that arise from a single primitive TIC which undergoes divergent or metaplastic differentiation.

Molecular evidence suggests that most uterine carcinosarcomas are monoclonal.^{15, 16, 17} Accordingly, supported by molecular, epidemiologic, genetic, and histologic data, uterine carcinosarcomas are now regarded as high-grade carcinomas with sarcomatous/stromal differentiation, similar to other organ systems.^{18, 19} This is reflected in the use of platinum based treatment regimens similar to those used for uterine carcinomas.

Why is it so interesting that uterine carcinosarcoma may be a uterine carcinoma?

If uterine carcinosarcomas are closely related to high-grade endometrial carcinomas, insight into the molecular pathogenesis of the latter can be exploited to identify therapeutic targets in the former. Knowledge of such targets provides an opportunity for a directed but comprehensive analysis of the genetics of carcinosarcomas with the possibility of identifying similar targets.

So what can we learn from endometrial cancer?

We are developing a better understanding of the molecular development of endometrial cancer. Increasingly both type I and type II endometrial carcinoma are being found to have genetic alterations affecting the AKT/PI3K pathway, a pathway which may be targeted by therapeutic agents.^20-25

Targeting the PI3K Pathway

Deletion of Phosphatase and tensin homologue on chromosome 10 (PTEN) is the most common genetic defect in endometrial cancer and is seen in up to 83% of tumours.^26-28 PTEN is part of the PI3k/AKT pathway which is an important signaling pathway in cancers. mTOR is a kinase which functions downstream of PI3K/Akt and inhibition of this is seen as an important therapeutic target in cancers.^29-31 Loss of PTEN causes deregulated PI3K/Akt/mTOR signalling which may provide neoplastic cells with a selective survival advantage by enhancing angiogenesis, protein translation and cell cycle progression.

Thus targeting the PI3K/MTOR pathway alone and in combination in endometrial cancer is being actively exploited as a treatment strategy in this disease.³⁶

Clinical Trials of Targeted Therapies in Endometrial Cancer

A recent clinical trial conducted by the National Cancer Institute of Canada (NCIC) investigating the MTOR inhibitor Temsirolimus, showed promising single agent activity in women with advanced and metastatic endometrial cancer.³⁷ Deforolimus, is an intravenous mTOR inhibitor and in a phase II trial for women with uterine malignancies (endometrial cancer and uterine carcinosarcoma); 28% of patients demonstrated either a complete or partial response to this drug.³⁸ Thus mTOR inhibitors do exhibit activity in women with advanced or recurrent endometrial cancer and carcinosarcoma.^39-46

Using Sequenom Mass ARRAY System to Analyze the PI3K Pathway in Uterine Carcinosarcoma

New sequencing and profiling technologies are allowing us to unlock the paraffin archive of tumor samples stored in hospital pathology departments.⁴⁷ One such technology is Sequenom matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) MassArray system.

In this project we will be analyzing mutation profile of the genes involved in the PI3K pathway in an annotated cohort of 70 uterine carcinosarcomas.

Significance

Uterine carcinosarcomas have poor prognosis and there are limited therapeutic options. This study will allow assessment of potential therapeutic targets in these cancers. Two recent publications provide strongly supportive data for this hypothesis. First, a group at Harvard has identified PIK3CA mutations in 19% of gynaecologic carcinosarcomas (uterine, ovarian, fallopian tube and vaginal) with mutations in the phosphatidylinositol 3-kinase pathway predominating in the uterine carcinosarcomas.⁴⁸ Second, another group has identified a 20% rate of PIK3CA mutations in uterine MMMTs using Sequenom technology.⁴⁹ This study will build on these observations by a comprehensive analysis of the PI3K pathway. If successful this will allow design of rational clinical trials for new treatments designed to target those abnormalities. In addition the study will contribute fundamental insights into the genetic alterations of these understudied tumors.

Grant Funding

The Liddy Shriver Sarcoma Initiative announced the funding of this $50,000 grant in June 2011. The study was made possible by a generous donation from Dr. Laura Somerville, by generous donations made in memory of Adrianne Pfeiffer (through several MMMT awareness events organized by her husband, Robert Pfeiffer), and by generous donations made in memory of Mary Anne Voeglie.